Oct. 5 , 19.51
S4T.T EFFECTS OF FAST REACTIONS I N
[ C O N T R I B U T I O N FROM THE
DEPARTMENT OF
CHEMISTRY AND CHEMICAL
4811
AQUEOUSSOT.UTIONS ENGINEERIXG, UNIVERSITY O F
CALIFORNIA]
Salt Effects on the Rates of Fast Reactions in Aqueous Solutions BY WILLIAME. HARTYAND G. K. ROLLEFSON RECEIVED APRIL 13, 1954 The effects of inert salts of various ionic types on the quenching of the fluorescence of the doubly positive ion of quinine by bromide ion and of the doubly negative ion of fluorescein by iodide ion have been studied. I t is found that the data for the effect of any one salt on a given reaction can be fitted by an equation derived with the aid of the Debye-Huckel theory. However, specific effects must be attributed to each salt in this theoretical treatment. The specific effects are much less in the quenching of fluorescein by iodide if the quenching constant is plotted against the positive ion concentration. It is also found that the quenching constant for the reaction of quinine ion with bromide can be expressed by a function determined by the concentrations and charges of the ions of the added salts but independent of their nature. It is pointed out that plots of this type may be of greater use for the correlation of kinetic data than the usual plots in terms of ionic strength.
It has been a common practice in the discussion of the effects of salts, which do not enter into the net reaction, on the rates of ionic reactions to plot the logarithm of the rate constant against the square root of the ionic strength of the solution. This type of plot is chosen because the Debye-Hiickel theory leads to the conclusion that such a plot will be linear and independent of the nature of the added salt a t low concentrations of such salts. In most systems data cannot be obtained a t sufficiently low concentrations t o fit the limiting law and the value of the rate constant a t zero ionic strength has to be obtained by a non-linear extrapolation. Recently Olson and Simonsonl pointed out that with some reactions between ions of like sign a more simple plot is obtained if the rate constant is considered to be a function of the concentration of the ions of opposite sign rather than the ionic strength. It has also been reported by Young and Jones2 that the ,Mayer3 equation for the activity coefficient of an ion leads to the conclusion that at high concentrations the value for a given ion depends much more on the charges and radii of the ions of opposite sign than it does on the properties of ions bearing the same sign. The present investigation was undertaken to test these points of view. The reactions chosen for study were the quenching of the fluorescence of the doubly negative fluorescein ion by iodide ion as an example of a reaction between ions of like sign and the quenching of the fluorescence of the doubly positive ion of quinine as an example of a reaction between ions of opposite sign. The quantity which we have measured is the quenching constant as defined by the SternVolrner equation Id1 = 1 ~ Q ( Q )
+
(1) A. R. Olson and T. R. Simonson, J . Chem. Phys., 17, 11676 (1949). Since the publication of this paper there has been diversity of opinion as to the origin of the idea. Daniels ( A n n . Rev. Phys. Chem., 1, 250 (1950)) speaks of it as a revolutionary idea. taidler ("Chemical Kinetics," McGraw-Hill Book Co., New York, N. Y , 1950. p. 127) says that evidently a new theoretical treatment of the behavior (of salt effects) is required. On the other hand Kilpatrick ( A n n . Rev. Phys. Chem., 2 , 270 (1951)) says that the idea was proposed in 1922 by Br@nsted. There is no doubt about the fact that Br@nsted(THISJOURNAL, 44, 898 (1922)) said that the activity coefficients of ions depend only on the action of ions of opposite sign and the salting out effect of the solvent. There is also no doubt that after the Debye-Huckel theory came out Brginsted as well as others interested in the effects of inert salts on rates invariably plotted their rate constants as a function of ionic strength. As far as we have been able to discover Olson and Simonson were the first to use the type of plots presented in their paper. (2) T. F. Young and A. C. Jones, A n n . Rev. Phys. Chem., S, 277
(1952). (3) J. E. Mayer, J . Chem. P h y s . , 18, 1426 (1950).
in which (Q) is the concentration of the added quencher. This constant is the product of the lifetime of the fluorescent substance and the bimolecular rate constant for the quenching process. The lifetime is not affected by the addition of inert salts hence the effects observed on the quenching constant can be assigned to the bimolecular rate constant, I n order to compare the absolute values of the rate constants for systems involving different fluorescent substances it is necessary to know the values for the lifetimes. Accurate measurements have been published recently4for both quinine and fluorescein so that it has been possible to convert all our observations to such terms. Reagents Used.-Eastman Kodak Co. fluorescein was dissolved in sodium hydroxide solution, the solution filtered and the fluorescein reprecipitated with acid. This process was repeated three times. The final product was dried to constant weight a t 110". The molar extinction coefficient of a 10-6 molar solution of this fluorescein in 0.001 36 sodium hydroxide was found to be 78.4 X lo3 a t 4950 A. The value in the literature is 78.1 X lo3. Potassium hydroxide solution was prepared by dissolving reagent grade compound in hot water until saturated, cooling to precipitate crystals of the hydrate, filtering, dissolving the crystals in water and diluting to a definite volume. The concentration of the solution was determined by titration with dried analytical grade potassium acid phthalate. Sodium hydroxide solution was prepared by dissolving reagent grade material in water until saturated, filtering the solution and diluting to the desired volume. The concentration was determined by the same method as for the potassium hydroxide. Merck U.S.P. quinine sulfate was recrystallized three times from water. The final product was dried to constant weight a t 110'. Quinine was obtained from some of this product by precipitation from water solution by sodium hgdroxide. After washing with hot water until no trace of sulfate could be detected in the filtrate, the quinine was dried to constant weight at 110'. The purified quinine melted a t 175-176" as compared t o the value in the literature of 175'. The absorption spectrum of the recrystallized quinine sulfate was identical to that of the quinine dissolved in the same concentration of sulfuric acid and in general agreement with the absorption spectrum given in the International Critical Tables. Sodium perchlorate solution was prepared by the neutralization of standardized perchloric acid with analytical reagent grade sodium carbonate, boiling off carbon dioxide and diluting to the desired volume. Lanthanum perchlorate was prepared by permitting lanthanum oxide to react with the required volume of standardized perchloric acid. The residue which failed to dissolve was filtered, washed and ignited. I t s weight corresponded to less than a tenth of a per cent. of the original oxide. The lanthanum oxide which was used was prepared by the ignition of C.P. lanthanum oxalate to constant weight at 1200'. A sample of the oxalate was titrated with (4) E. A. Bailey, Jr., and G. K. Rollefson, ibid., 21, 1315 (1953).
WILLIAME. HARTYAND G. K. ROLLEFSON
4512
standardized permanganate and the available lanthanum oxide calculated. The calculated value agreed within less than 0.2% with the weight found after ignition. Magnesium perchlorate was prepared by the reaction of reagent grade magnesium oxide, which had been ignited to constant weight at 1200", with the required volume of standardized perchloric acid. Sodium pyrophosphate was prepared from C.P. potassium pyrophosphate b y precipitation with sodium acetate. The precipitate was recrystallized from water three times and dried to constant weight a t 110'. Qualitative tests for potassium ion and phosphate ion in the final product were negative. The potassium iodide, potassium nitrate, potassium sulfate, potassium citrate, potassium acetate, potassium bromide and sodium sulfate used were reagent grade. They were dried t o constant weight and then used without further purification. Apparatus.-Figure 1 is a schematic diagram of the apparatus used t o measure the light intensities. Light from the General Electric A H 4 mercury vapor lamp X was collimated by the lens L, passed through the filter Fl of Yo. 5860 Corning glass (this -transmits the mercury lines around 3660 A,), and the beam split by the filter M of No. 7910 Corning glass. One part of the beam passed through the assembly Fz which consisted of a solution of the fluorescent substance being used in the given set of experiments followed by a Corning 3060 filter which absorbs the ultraviolet light but transmits the fluorescent light from the solution. The other part of the beam passed into the cell C which was set in a water thermostat and contained the fluorescent solution under investigation. The fluorescence from this solution passed through the Corning 3060 filter Fa t o the photocell Pz. The photocells PIand P? were U'eston Photronic Model 594 barrier layer type photocells. I- - - - - - - - - - - - - -. I
1
!
-
I
!I
IF3
i n
X
Fig. 1.-Diagram
of apparatus.
In any experiment the cell C was one of a set of 2.5 cm. 0.d. Pyrex tubes which were made from the same piece of Pyrex stock so that their optical properties would be as nearly alike as possible. The cell, with the fluorescent solution in it, was clamped rigidly in place in the water-bath which was equipped with the devices necessary t o control the temperature within a tenth of a degree. The waterbath was a 4.5-liter monel vessel fitted with Corning 7910 windows. The light intensities were measured by balancing the current from photocell P? against that from Pi by the arrangement shown in the diagram. By this arrangement the intensity of the fluorescent light from C was always determined relative to the intensity of the exciting beam. Hence any effects due to fluctuations in the intensity of the exciting light were cancelled out. Both photocells were cooled with jets of compressed air t o eliminate effects due to temperature fluctuations I n order t o get the best reproducibility of readings it was found desirable to use peg type resistance for RI, RBand Ra. The galvanometer was a Leeds and Northrup No. 2430 with a resistance of 24 ohms and a sensitivity of 0.003 microamp. per m m . The resistance Rc was set at the critical damping resistance of 450 ohms. I n making the measurements the usual practice was to set the resistance R3at 500 ohms and the sum of RI and Ro a t 10,000 ohms. Under these conditions, with both cells illuminated and the circuit balanced, the reading of RI was
Vol. 76
proportional to the intensity of the light falling on P2. Tests were made with screens and filters to check the linearity of this scale. Under typical operating conditions the bridge was easily balanced to 0.1% of the resistance Rp. The typical experimental procedure was t o make consecutive measurements on a set of five sample tubes until a t least three sets of readings agreeing to 0.1% or better had been obtained. One of the tubes contained distilled water and served as a measure of the blank due to scattered light which had not been removed by the filters. The other four were divided into two sets of two. One of each set was filled with the fluorescent solution which contained quencher and the other was filled with a solution which \vas identical with the first except that it contained no quencher. The correction for the distilled water blank amounted to 0.1-0.2% of the intensity of the fluorescence. I n order to correct for any differences between tubes the two sets were filled with itlentical solutions and the intensities meawred . This correction ranged from 0.1-0.3'%. After making these corrections the values of I o / I were calculated for each quenched solution. Values for solutions of the same composition but prepared a t different times were found t o agree within 0.2%. Under the usual experimental conditions the corresponding error in the determination of the quenching constant was about 0.5y0. Experimental Results.-All measurements with quinine were made in 0.01 molal perchloric acid and those with fluorescein in 0.001 molal potassium hydroxide to insure having all of the fluorescer in the form of the divalent ion. I n the experiments with fluorescein in which the added salt was sodium pyrophosphate, the hydroxide concentratioii was raised t o 0.005 molal to ensure having the pyrophosphate in the form of the quadrivalent ion. 111 all experiments the fluorescer was present a t a concentration of 10-4 ,If. I n the work with quinine the potassium bromide was 0.001 Ad and in that with fluorescein the potassium iodide was 0.02 M except for one series of experiments in which the iodide was varied from 0.02 to 0.002 d l with no inert electrolyte present. Such concentrations gave approximately 25% quenching, which is sufficient to provide good accuracy in the determination of the quenching constants but avoids deviations from the Stern-Volmer law. All measurements were at. 25' unless otherwise indicated. The salts added t o the quinine-brornide system were all perchlorates. In the fluorescein-iodide system they were aiways either the potassium or sodium salts. Hence in the tables of data the column headed normality of added salt is always the concentration in moles per liter of the perchlorate ion (or potassium or sodium ion, as thc case may be) introduced by the added salt. The data for the quenching of quinine by bromide are presented in Table I and those for fluorescein by iodide in Tables I1 and 111. (For the sake of brevity, if a reading at ii given normality was taken for only one salt it has been omitted from the table, actually some additional readings wrre obtained for each of the salts.) The ionic strengths of these solutions are 0.01 or higher. This minimum ionic strength is fixed by the fact that in the quinine-bromide reaction the perchloric acid concentration had to be kept a t 0.01 -If, while in the fluorescein-iodide reaction lowering the
TABLE I THEQCENCHINC
OF
B Y BROMIDEION PERCHLORIC ACID
QUININEION
IN
0.01
;lr
k, liters/rnole NaClO,
h.Ig(CIOdz
La(ClO4)s
Added salt, A'
347.0
347.0 295.5 268.4 249.4 233.1 222.2 212.9 300.4 187.8 179.1 172.1 164.5 152.0
347.0 288.3 260.0 340.7 22,5,3 214.9 202.8 190.3 179.6 171.6 164.9 158.5 147.0
0.0000 .0100 .0200 ,0300 ,0400 .0500 .0600 ,0800
:3n6 . 9
279.7 259.3 247.0 236, n 225.3 212.0 199.9 191.1 181.6 174.8
164.3
.loo0 ,1200 ,1400 ,1600 ,2000
Oct. 5 , 1954
SALT
EFFECTS OF FAST REACTIONS IN AQUEOUSSOLUTIONS
potassium iodide concentration below 0.02 M reduced the accuracy of the measured values of the quenching constant. Some experiments were run with the latter reaction at lower ionic strength by reducing the iodide concentration and hav-
TABLE I1 THE QUENCHING O F FLUORESCEIN I O N BY IODIDE 0.001 M POTASSIUM HYDROXIDE
.
k , liters/molc KCzHaOi KzSO,
KNOs
10.04 10.80 11.18 12.11 12.76 13.43 14.17 14 67 15.65
10.04 11.43 12.19 12.96 13.52 14.47 15.16
Added salt,
N
KsCsHrO7
10.04 10.90 11.33 12.26 12.90 13.67 14.61 15.17 16.39
0.0000 .0100 .0200 ,0400 ,0600 .0800 .1200 .1600 .2400
10.04 10.95 11.39 12.30 12.92 13.70 14.58 15.46 16.55
TABLE 111 THE QUENCHING O F FLUORESCEIN ION BY IODIDE 0.005 hf SODIUM HYDROXIDE k, liters/mole
NaClOr
10.34 11.29 12.25 12.80 13.24 14.03 14.70 15.04
ION I N
ION I N
NazSOa
PiarPzOi
Added salt, N
10.34 11.57 12.36 12.99 13.55 14.37 15.09 15.52
10.34 11.26 12.05 12.66 13.22 14.04
0.0000 .0200 ,0400 .0600 .0800 .1200 ,1600 ,2000
... 15.04
TABLE IV THE QUENCHINGOF FLUORESCEIN ION BY IODIDE 0.001 M POTASSIUM HYDROXIDE k, liters/mole
Concn. of KI
10.08 9.59 8.94 8.82 8.65 8.25 7.53
0.0200 ,0150 ,0100 .0080 .0060 .0040 ,0020
ION IN
4813
ing no other added salt. These results are given in Table IV. Table V shows the effect of temperature on these reactions. The apparent discrepancy between the quenching constants for fluorescein with iodide a t 25‘ as given in this Table and in Table I1 is caused by a change in the position of the mercury lamp before the data on the temperature effect were taken. The observed quenching constant in this case is somewhat dependent on the geometry.
Discussion First, we shall compare our results with the equations derivable from the Debye-Huckel theory. It is apparent from an inspection of the data presented that we are dealing with ionic strengths beyond the range of the limiting law form of that theory. The next approximation leads by the usual methods to the following equation for the rate constant in aqueous solutions a t 2506 log k = log ko + 1.018z~za/(Bf l/p‘/z) Strictly speaking, this equation applies only to the bimolecular rate constant, but since it has been found that the lifetime of the fluorescent substance is not dependent on the concentration of the inert salts present we may apply the equation to the experimental values of the quenching constant. Since K O is the value a t zero ionic strength it should be the same for all solutions involving the same fluorescer and quencher. I n testing the equation we have treated ko as a possible variable since the derivation of the above equation assumes the theoretical value for the limiting slope and we did not wish to introduce any other fixed values. A graphical test of the equation is obtained by plotting log k against the reciprocal of B l/j.~’/~. A typical plot of this kind is shown in Fig. 2. It is apparent that the data fit the straight line called for by the equation. Similar lines were obtained for the data with the other added salts. In some cases the point a t the lowest ionic strength was off the line which may be because of experimental error or may be caused by the fact that B is dependent on the nature of the
+
TABLE V TEMPERATURE DEPENDENCE O F QUININE-BROMIDE REACTION
Added salt, Added salt
N
25’
......
0.0000 .0200 .OS00 .1600 .0100 .0400 .0800
347.0 279.7 212.0 174.8 289.3 225.3 100.3
NaCIOI
La( ClOJp
k, liters/mole 35O
45O
56‘
422.0 338.4 253.6 214.1 354.3 276.6 227.4
494.4 398.3 304.1 253.9 406.5 319.2 272.8
575.3 467.2 344.3 291 3 479.3 380.8 316.9
TEMPERATURE DEPENDENCE O F FLUORESCEIN-IODIDE REACTION
k, liters/mole
Added salt
Added salt,
N
5’
15’
...
0.0000 .0200 .0900 .lC;OO .2400 .0600 .IO82 .1600
6.85 7.52 9.14 9.95 10.45 8.58 9.35 10.00
8.33 9.34 11.13 12.07 12.83 10.67 11.78 12.51
KNOI
&so4
25O
35O
45O
10.20 11.42 13.63 14.78 15.87 13.15 14.39 15.39
11.73 13.17 15.74 17.23 18.26 15.01 16.45 17.74
12.96
14.53 17.65 19.29 20.52 16.80 18.51 19.75
- I
Fig. 2.-Effect -.
I B t J Z
of potassium nitrate on the rate of the fluorescein-iodide reaction.
( 5 ) Cf A. A. Frost and R . G Pearson, “Rlnctics and Mechanism,” John Wiley and Sons, Inc., New York, N. Y., 1953, p. 138.
4814
I
r E. ~HARTY ~ AND ~ ~G . K. ~ KOLLEFSON ~ ~
A
Vol. 76
salt making up the ionic strength. At high ionic strengths the added salt is the dominating factor in determining the value of B but for low concentrations some deviation may be found because of the fact that under such conditions the fluorescer and quencher are dominant in determining the properties of the solution. Least squares were used to determine the best values of log ko and B for each added salt and the values are listed in Table VI. In the case of potassium citrate as the added salt the values given are for the best straight line fitting the data although in that case the data could be fitted better with a curve. Inspection of the Table shows that the variation in log ka is negligible, especially when consideration is given to the amount the data have to be extrapolated to fix this value. The values of B can be varied slightly from the ones given but we believed that such a change could not be greater than about 5y0without showing a noticeable deviation of the data froin the theoretical curve. TABLE1.1 DEBYE-H~CKEL THEORY
PARAMETERS C S E D I N T E S T O F Added salt B
3.24 3.00 3.26 :3,43 2.70 t i . 37
KSO3 KCzHjOi
KzSO1 K:sCsHaOi KI XaClOa Na2S04 SaaP?O: SaCIOI Mg( ClO,)? La( Clod3
:1 .38
''a,''
9.86 9.13 9.92 10 41 8.28 9 9.5
I! I , 29 1 I .08 5.72 f i L'T t i . 15
TtiR 8.717 2.721 2.728
:1, 61
1 .X8 2.06 2.12
i.
log ko
0.804 ,800 ,790 .785 ,790 so2 ,793
The previous treatment indicates that our data are compatible with the usual method of applying the theory of Debye and Hiickel to data of this kind. I n order to apply the ideas of Olson and Simonson to our data we have made the plot of k "LIS.concentration of positive ion shown in Fig. 3 . It is apparent
0 NaC104
028
-
,020
-
p011
A
8 0
2-
-
-
0
m G
U
a
0.12
o m
L'J(C104)3
8
so,,
!
W(C1042
0 0.24
0-KNO3
a L
0 A
A
A
-
8
0 008
E 009
" a
B
-
B c. 0.04
A%
-
Ro
0 05
0
A .
, 110
120
140
13.0
I
150
d
160
k,
Fig. 3.--Plot of qucnchitig co~istantfor iodide 011 fluorcscciii against thc positive ioii coiicentratioii.
I
I
I
I
1
110
200
230
260
290
0
,
Oct. 5, 1954
PHOTOLYSIS
OF
METHYLC Y C L O P R O P Y L KETONE
4815
+
be plotted against P ' / ~ + 3p1/%- rather than decreased yield can be caused by the lifetime being against the ordinary ionic strength. Such a plot is decreased by a n increase in the rates of the competshown in Fig. 4. It is apparent that the data can be ing first order reactions. It is possible now to measfitted within the limits of experimental error by a ure lifetimes of solutions such as these with suffisingle curve. For the present this method of plot- cient accuracy to test such an explanation; therefore, ting the data must be considered as purely empirical we shall defer further discussion of the temperature although i t was suggested by some theoretical con- effects until such measurements are available. The conclusions which may be drawn from the siderations. The analysis of the effect of temperature on the data we have presented are not as definite as we quenching process is complicated by the fact that had hoped that they might be. Certainly the fit the quenching constant is the product of the life- with the straight line based on the Debye-Huckel time of the excited molecule and a bimolecular rate theory, which is shown in Fig. 2, must be taken as constant. If the lifetime is constant the entire tem- support for the equation based on that theory, alperature effect is that on the bimolecular rate con- though it does not necessarily confirm the physical stant. Then a straight line should be obtained if the interpretation put on the parameters. On the logarithm of the quenching constant is plotted other hand, the curves shown in Fig. 3 and Fig. 4 against the reciprocal of the absolute temperature. indicate that specific effects are less pronounced if We found that with both the systems studied in stress is laid on the effect of ions of opposite sign this paper such a plot yields a curved line. Such a to those under consideration. On the basis of our curvature could be caused by a variation of the life- experience we would say that rate constants detertime with temperature. Actually, Lewschin6has re- mined in solutions of variable ionic constitution can ported that the fluorescence yield from solutions of be correlated best by plots of the type shown in fluorescein decreases as the temperature is raised. A the latter figures. BERKELEY.CALIFORNIA (6) W. L . Lewschin, Z . physik, 43, 230 (1927).
[CONTRIBUTION FROM THE CHEMICAL
LABORATORIES, NORTHWESTERN UNIVERSITY]
Structure and Reactivity in the Vapor-Phase Photolysis of Ketones. Cyclopropyl Ketone1p2 B Y J. N.
PITTS,
JR., AND
I. Methyl
I. N O R M A N
RECEIVED MARCH24, 1954 The vapor-phase photolysis of methyl cyclopropyl ketone a t 2654-2537 A. leads primarily to rearrangement giving methyl propenyl ketone with a quantum yield of 0.31 zk 0.02 from 25 to 120'. The probability of dissociation of the parent ketone into radicals must be small, since the quantum yield of carbon monoxide is only 0.12 a t 170'. Minor non-condensable products are 1-butene, propylene, ethane, methane, ethylene and biallyl and, for the runs above loo", traces of cyclopropane. A reaction sequence is proposed that accounts for the olefinic nature of the non-condensable products, and the formation of the a,p-unsaturated ketone.
I n recent years i t has become increasingly evident that the structures of alkyl substituents have a pronounced effect upon the primary modes of photodecomposition o€ simple aliphatic ketone^.^ I t is now recognized that all simple aliphatic ketones photodissociate to some extent by a Norrish Type I free radical process4 to give alkyl and acyl radicals. Acetone, 3h,5 methyl ethyl and diethyl ketoneg decompose almost exclusively in this manner. However, ketones with larger alkyl groups, such as methyl n-propyl ketone,3b di-n(1) Presented a t the Los Angeles Meeting of the American Chemical Society, March, 1953. (2) Taken from t h e doctoral dissertation of I. Norman, Northwestern University, 1953. (3) See t h e comprehensive reviews, (a) W. Davis, Jr., Chem. Reus., 40, 201 (1947); (b) A. J. C. Nicholson, Rev. P u r e and APPlied Chem., 1, 174 (1952). (4) R. G. W. Norrish a n d M. E. S. Appleyard. 1.Chem. SOC.,874 (1934). (5) W. A. Noyes, Jr., a n d L. M. Dorfman, J . Chem. P h y s . , 16, 788
(1948). (6) V. R. Ells and W. A. Noyes, Jr., THIS J O U R N A L , 61, 2492 (1939). (7) W. J. Moore and H. S. Taylor, J . Chem. Phys., 8 , 466 (1940). (8) J. N. P i t t s , Jr., a n d F. E. Blacet, THIS J O U R N A L , 71, 2810 (1950). (9) K. 0. Kutschke, M. J. H. Wijnen a n d E. W. I
